CN107995600A - A kind of wireless sense network collecting method based on matrix completion - Google Patents
A kind of wireless sense network collecting method based on matrix completion Download PDFInfo
- Publication number
- CN107995600A CN107995600A CN201711138648.5A CN201711138648A CN107995600A CN 107995600 A CN107995600 A CN 107995600A CN 201711138648 A CN201711138648 A CN 201711138648A CN 107995600 A CN107995600 A CN 107995600A
- Authority
- CN
- China
- Prior art keywords
- data
- node
- matrix
- base station
- completion
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W84/00—Network topologies
- H04W84/18—Self-organising networks, e.g. ad-hoc networks or sensor networks
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The invention discloses a kind of wireless sense network collecting method based on matrix completion.It is broadly divided into sensing node and the link of base station two.In sensing node, one wheel data of sensing node collection, RLS adaptive filtering algorithms are recycled to estimate the reading at the moment, if the error amount of actual perceived data and estimated data exceedes certain upper limit, the information such as node serial number, collection moment, gathered data are then transferred to base station, otherwise do not upload data.In a base station, base station perceives the data of node-node transmission by receiving, builds an incomplete gathered data matrix B, then carry out completion, the matrix B after being restored using matrix completion restructing algorithm to it ', which stores complete perception data.The present invention can efficiently reduce the energy consumption of sensor node, improve sampling end work efficiency, extend the life cycle of whole network.
Description
Technical field
The invention belongs to wireless sensor network technology field, is related to a kind of wireless sense network based on matrix completion technology
Data acquisition technology, is mainly used in wireless sensor network, has the data acquisition amount that reduces, reduces node energy consumption, extends
Network lifecycle etc. acts on.
Background technology
Wireless sensor network (Wireless Sensor Networks, WSNs) is a kind of special Ad-hoc networks,
By having the function of that the sensor node of sensing, data processing and short-distance wireless communication forms, it is not required fixed network to support,
With the features such as rapid deployment, survivability is strong, can be widely applied to military and national defense, environmental monitoring, biologic medical, intelligent transportation,
The field such as rescue and relief work and business application.
The gathered data of wireless sensor network both is from sensing node.Sensing node is generally battery powered, and more
Change and be not easy.In sensor network, the collection and transmission of data can all cause the consumption of electric energy.In order to reduce energy consumption, extend net
Network life cycle, can take some technological means to reduce data volume.Wherein, data suppress in net and compression is to reduce data
The common thinking of amount, it can fundamentally reduce energy consumption and extend Network morals.This kind of method includes tradition
Message sink coding, distributed source coding (Distributed Source Coding, DSC), compressed sensing (Compressive
Sensing, CS) etc..Message sink coding is data suppressing method in a kind of net using coding staff data space correlation, in order to
More preferable compression effectiveness is obtained, it usually requires mutually coordinated between sensing node.Distributed source coding is message sink coding
A kind of optimization, it attempts to reduce the complexity of data in sensing node, and make use of spatial coherence in a base station.Compression
Perception is a kind of method for seeking sparse solution to deficient alignment sexual system.But these methods suffer from the drawbacks of similar, need
Higher time space complexity is spent, and relatively depends on the correlation of adjacent node.
Matrix completion (Matrix Completion, MC) can solve the above problems well.Matrix completion is a profit
Recover the technology of the loss entry in matrix with low-rank feature.As the extension of compressed sensing, it has been applied to respectively
In kind research field.And approximate low-rank characteristic possessed by wireless sense network perception data so that matrix complementing method is very suitable
Collection for data.As long as based on some a small amount of perception datas, matrix completion is with regard to that can recover whole readings.Therefore,
Synchronization only needs part of nodes to submit data.The present more existing data acquisition side based on matrix completion technology
Case.But in existing scheme, majority is simply according to certain probability random acquisition data, less pertinence, so as to cause matrix
The precision for recovering data after completion is relatively low.
The content of the invention
The present invention is directed to the low-rank characteristic of sensor network perception data, and using matrix completion algorithm, and it is adaptive to combine RLS
Filter algorithm is answered, a kind of collecting method based on matrix completion is proposed, for solving traditional acquisition scheme high energy consumption, network
The problems such as life cycle is short.This programme can reduce data traffic, reduction communication is prolonged while gathered data quality is ensured
Late, and Network morals are extended.
To achieve the above object, the technical solution adopted by the present invention is a kind of wireless sensing network data based on matrix completion
Acquisition method, is cooperated and is completed by node and the part of base station two, node section specifically comprises the steps of:
Step A1:Node initializing, the data before removing, set evaluated error θmaxWith sampling instant t;
Step A2:T=0, sensing node collection first run data, each node is by its node serial number i, collection moment t and sense
Primary data Xi(t) it is sent to base-station node;
Step A3:T=t+1, sensing node continue the wheel data of collection one Xi(t), and preserve;
Step A4:Node estimates the reading C of t moment using RLS adaptive filterings algorithmi(t);
Step A5:Node carries out adaptive filtering assessment to data, if actual perceived data Xi(t) with estimated data Ci
(t) error amount is more than θmax, then by Xi(t), i, t are transferred to base station, otherwise do not upload data;
Step A6:Such as acquisition tasks completion, then node is stopped, and otherwise, is transferred to step A3;
Base station portion specifically comprises the steps of:
Step B1:System initialization, data before are emptied, and set network node sum K, and sampling processing cycle
T;
Step B2:Structure one K × T two-dimensional matrix B come memory node submission perception data, matrix element it is initial
Value is set to NaN;
Step B3:The data of receiving node transmission;
Step B4:Received data are docked, according to node serial number i, time t and transmission data Xi(t), from initial matrix B
In find corresponding position, and use Xi(t) NaN is replaced;
Step B5:Repeat step B3 and B4, until a cycle T completions, at this moment obtain an incomplete gathered data
Matrix B, houses the perception data after RLS algorithm filters in B;
Step B6:Base station carries out completion, the square after being restored using matrix completion restructing algorithm to perception data matrix B
Battle array B ', the matrix store complete perception data;
Step B7:Perception data after recovery is submitted to data processing centre by base station.
Further, above-mentioned each node possesses unique ID numberings, and sampling node stores previous sampled data, uses
Calculated in the valuation of adaptive filtering next time.
Further, above-mentioned sensing node can be predicted to obtain a desired signal by RLS adaptive filter, only current
Signal is hoped to differ by more than certain error amount with actual result and can just transmit the data.
Compared with prior art, beneficial effects of the present invention:
(1) compared with traditional data acquisition plan, the present invention at sensing node using RLS adaptive filterings algorithm into
Row filtering, only when predicted value and actual read number differ by more than certain limit, can just transmit the data.So can effectively it subtract
Data transfer in few network, so as to play the role of reducing energy consumption, extend network lifecycle.
(2) compared with traditional data acquisition plan, the high calculating of complexity is transferred to by the present invention from sensing node
In base station, so as to reduce the calculating energy consumption of node, the working life of sensing node is extended.
(3) compared with traditional data acquisition plan, when the present invention carries out data filtering at sensing node, to data mistake
The error of filter is controlled, this make it that the accuracy rate for finally recovering data is higher than the accuracy rate of general data acquisition scheme.
Brief description of the drawings
Fig. 1 is the work flow diagram of sensing node.
Fig. 2 is the work flow diagram of base station portion.
Embodiment
In conjunction with attached drawing, the present invention is described in further detail.
The present invention proposes a scheme for allowing the adaptive gathered data of sensing node, can be with of a relatively high accuracy resumption
Go out the reading of missing.Present invention utilizes the low-rank characteristic of perception data, matrix completion technology and RLS (Recursive are employed
Least Square, least square method) adaptive filtering algorithm.The present invention utilizes the adaptive mistakes of RLS first at sensing node
Algorithm is filtered to carry out the prediction of data and filtering, this can be just transmitted when gathered data differs by more than a certain range with prediction result
Data.Therefore sensing node need not submit all data.In base station end, base-station node is mended according to received data into row matrix
Entirely, to recover whole readings.So not only the higher calculating of complexity has been transferred in base station at node, while
Reduce the transmission and transmission of data, compared with traditional scheme, the transmission quantity of data is smaller in the present invention, can reduce energy
Consumption and bandwidth occupancy, can preferably extend network lifecycle.Assuming that deploying K in the target area perceives section
Point, each node have unique numbering i.The invention mainly comprises sensing node and the part of base station two.
First, the workflow of sensing node part is as follows, its flow chart is as shown in Figure 1.
1. node initializing, the data before removing, set evaluated error θmaxWith sampling instant t;
2.t=0, sensing node collection first run data.Each node is by its node serial number i, collection moment t and perception data
Xi(t) it is sent to base-station node;
3.t=t+1, sensing node continue the wheel data of collection one Xi(t), and preserve;
4. node carries out the prediction of data using RLS adaptive filterings algorithm, the estimated value C of t moment is obtainedi(t);
5. node carries out adaptive filtering to data.If actual perceived data Xi(t) with estimated data Ci(t) error
Value is more than θmax, that is, work as | Xi(t)-Ci(t)|>θmaxWhen, then by Xi(t), i, t are transferred to base station, otherwise do not upload data;
6. such as acquisition tasks completion, then node is stopped.Otherwise, 3 are transferred to.
2nd, the workflow of base station portion is as follows, its flow chart is as shown in Figure 2.
1. system initialization, data before are emptied.Sampling processing cycle T is set.
2. in a base station, build the perception number that K × T two-dimensional matrixes B as shown in formula (1) carrys out memory node submission
According to wherein bi(j) what is represented is at the j moment, and the perception data for the node that numbering is i, the initial value of matrix element is set to
NaN, represents null value.
3. base station receives the submission data from node, including node serial number i, collection moment t and perception data Xi(t)。
4. base-station node is according to node serial number i, timing node t and transmission data Xi(t), found from initial matrix B pair
The value b answeredi(t), and X is usedi(t) it is replaced.
5. repeating 3~4 until a sampling period T completion, an incomplete gathered data matrix has thus been obtained
B, what is stored in the matrix is the perception data after RLS algorithm filters.
6. completion, the complete matrix B after being restored are carried out to B using restructing algorithm ', which stores completely
Perception data.
Specifically, the matrix reconstruction algorithm that uses of the present invention be accurate method of Lagrange multipliers (exact ALM,
EALM)., can be by solving following order for incomplete gathered data matrix B due to the characteristic of perception data approximation low-rank
Minimization problem realizes the completion to unknown element:
Wherein B and B ' is the matrix of K × T dimensions, and B is incomplete gathered data matrix, and B ' is then reconstruct to be asked
Matrix, Ω are by the location sets of sampling element.In order to recover the low-rank structure of sampling matrix B, sampling matrix B can be decomposed into
The sum of two matrixes, i.e. B=M+N, wherein matrix M and N are unknown, but M is low-rank, when the element Gaussian distributed of matrix N,
It can solve to obtain optimal matrix M with classical Principal Component Analysis.Singular value decomposition is carried out to sampling matrix B again, works as calculation
Method result after time iteration enough restrains, and can obtain the gathered data matrix B of energy approximate representation initial data '.
7. base station is by complete perception data matrix B ' it is submitted to data processing centre.
Embodiment
In order to facilitate description, one embodiment is now provided:Assuming that need to develop now a kind of based on the wireless of matrix completion
Sensor Network data acquisition technology, the primary demand of application are to reduce data traffic, extend the life cycle of wireless network, gather number
According to accurate.Acquisition scheme includes sensing node and the part of base station two, as follows respectively.
First, sensing node part
Step 1:Node initializing;
Step 2:Initial time, sampling instant are set to 0, and sensing node is sampled and sends the data to base station;
Step 3:Sampling instant adds 1, and sensing node continues the wheel data of collection one and preserves;
Step 4:Node obtains estimated data using RLS adaptive filterings algorithm to carry out the prediction of data;
Step 5:Node carries out adaptive filtering assessment to data, when actual perceived to the obtained data of data and prediction
Data are then transmitted to base station by error amount more than certain value, otherwise do not upload data;
Step 6:Acquisition tasks are completed, then node is stopped.Otherwise, it is transferred to step 3.
2nd, base station portion
Step 1:System initialization;
Step 2:Matrix B is built to store the perception data of network submission;
Step 3:The data of receiving node transmission;
Step 4:According to received data update matrix B;
Step 5:Repeat step 3 is completed to step 4 until a sampling period, has obtained gathered data matrix B;
Step 6:Base station carries out completion, the matrix B after being restored using matrix completion restructing algorithm to B ';
Step 7:Complete matrix after recovery is submitted to data processing centre by base station.
The present invention is transferred to base station from sampling end by introducing matrix completion algorithm, by a large amount of evaluation works and carries out, and only needs
Compressed data can be obtained by carrying out fractional-sample to initial data.This method can effectively reduce the energy consumption of sensor node,
Sampling end work efficiency is improved, extends the life cycle of whole network.
Claims (3)
1. a kind of wireless sense network collecting method based on matrix completion, has been cooperated by node and the part of base station two
Into, it is characterised in that
Node section specifically comprises the steps of:
Step A1:Node initializing, the data before removing, set evaluated error θmaxWith sampling instant t;
Step A2:T=0, sensing node collection first run data, each node is by its node serial number i, collection moment t and perceives number
According to Xi(t) it is sent to base-station node;
Step A3:T=t+1, sensing node continue the wheel data of collection one Xi(t), and preserve;
Step A4:Node estimates the reading C of t moment using RLS adaptive filterings algorithmi(t);
Step A5:Node carries out adaptive filtering assessment to data, if actual perceived data Xi(t) with estimated data Ci(t)
Error amount is more than θmax, then by Xi(t), i, t are transferred to base station, otherwise do not upload data;
Step A6:Such as acquisition tasks completion, then node is stopped, and otherwise, is transferred to step A3;
Base station portion specifically comprises the steps of:
Step B1:System initialization, data before are emptied, and set network node sum K, and sampling processing cycle T;
Step B2:The two-dimensional matrix B of one K × T of structure carrys out the perception data of memory node submission, and the initial value of matrix element is equal
It is set to NaN;
Step B3:The data of receiving node transmission;
Step B4:Received data are docked, according to node serial number i, time t and transmission data Xi(t), found from initial matrix B
Corresponding position, and use Xi(t) NaN is replaced;
Step B5:Repeat step B3 and B4, until a cycle T completions, at this moment obtain an incomplete gathered data matrix
The perception data after RLS algorithm filters is housed in B, B;
Step B6:Base station carries out completion, the matrix after being restored using matrix completion restructing algorithm to perception data matrix B
B ', the matrix store complete perception data;
Step B7:Perception data after recovery is submitted to data processing centre by base station.
2. the wireless sense network collecting method based on matrix completion according to right 1, it is characterised in that:Each node
Possess unique ID numberings, sampling node stores previous sampled data, based on the valuation of adaptive filtering next time
Calculate.
3. the wireless sense network collecting method based on matrix completion according to right 1, it is characterised in that:Sensing node
It can predict to obtain a desired signal by RLS adaptive filter, only when desired signal and actual result differ by more than
Certain error amount can just transmit the data.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711138648.5A CN107995600A (en) | 2017-11-16 | 2017-11-16 | A kind of wireless sense network collecting method based on matrix completion |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201711138648.5A CN107995600A (en) | 2017-11-16 | 2017-11-16 | A kind of wireless sense network collecting method based on matrix completion |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107995600A true CN107995600A (en) | 2018-05-04 |
Family
ID=62031536
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201711138648.5A Pending CN107995600A (en) | 2017-11-16 | 2017-11-16 | A kind of wireless sense network collecting method based on matrix completion |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107995600A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109460398A (en) * | 2018-10-12 | 2019-03-12 | 深圳绿米联创科技有限公司 | Complementing method, device and the electronic equipment of time series data |
CN110505657A (en) * | 2018-05-16 | 2019-11-26 | 中南大学 | Method of data capture based on matrix fill-in technology in a kind of wireless sensor network |
CN111954305A (en) * | 2020-07-08 | 2020-11-17 | 电子科技大学 | Wireless sensor node task time slot scheduling method based on matrix completion |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103139804A (en) * | 2013-02-07 | 2013-06-05 | 西安邮电大学 | Energy-saving transmission self-adaption recursive least squares (RLS) distributed-type detection method of wireless sensor network |
CN103598873A (en) * | 2013-09-27 | 2014-02-26 | 四川大学 | Physiological signal intelligent monitoring system based on self-adaptation wireless sensor network |
CN105871762A (en) * | 2016-05-23 | 2016-08-17 | 苏州大学 | Adaptive network used for estimation of sparse parameter vector |
US20160284003A1 (en) * | 2015-03-26 | 2016-09-29 | Xerox Corporation | Time-sensitive collaborative filtering through adaptive matrix completion |
CN107205154A (en) * | 2017-06-07 | 2017-09-26 | 南京邮电大学 | A kind of radio multimedia sensor network compression of images acquisition method based on matrix completion |
-
2017
- 2017-11-16 CN CN201711138648.5A patent/CN107995600A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103139804A (en) * | 2013-02-07 | 2013-06-05 | 西安邮电大学 | Energy-saving transmission self-adaption recursive least squares (RLS) distributed-type detection method of wireless sensor network |
CN103598873A (en) * | 2013-09-27 | 2014-02-26 | 四川大学 | Physiological signal intelligent monitoring system based on self-adaptation wireless sensor network |
US20160284003A1 (en) * | 2015-03-26 | 2016-09-29 | Xerox Corporation | Time-sensitive collaborative filtering through adaptive matrix completion |
CN105871762A (en) * | 2016-05-23 | 2016-08-17 | 苏州大学 | Adaptive network used for estimation of sparse parameter vector |
CN107205154A (en) * | 2017-06-07 | 2017-09-26 | 南京邮电大学 | A kind of radio multimedia sensor network compression of images acquisition method based on matrix completion |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110505657A (en) * | 2018-05-16 | 2019-11-26 | 中南大学 | Method of data capture based on matrix fill-in technology in a kind of wireless sensor network |
CN109460398A (en) * | 2018-10-12 | 2019-03-12 | 深圳绿米联创科技有限公司 | Complementing method, device and the electronic equipment of time series data |
CN111954305A (en) * | 2020-07-08 | 2020-11-17 | 电子科技大学 | Wireless sensor node task time slot scheduling method based on matrix completion |
CN111954305B (en) * | 2020-07-08 | 2022-11-15 | 电子科技大学 | Wireless sensor node task time slot scheduling method based on matrix completion |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Wu et al. | Data prediction, compression, and recovery in clustered wireless sensor networks for environmental monitoring applications | |
CN106341842B (en) | Method and device for transmitting wireless sensor network data | |
CN101848529B (en) | Method for compressing multiple principle component analysis data of wireless sensor network | |
Larios et al. | Energy efficient wireless sensor network communications based on computational intelligent data fusion for environmental monitoring | |
Tayeh et al. | A spatial-temporal correlation approach for data reduction in cluster-based sensor networks | |
CN107995600A (en) | A kind of wireless sense network collecting method based on matrix completion | |
CN103280084B (en) | A kind of collecting method of multi-parameters real-time monitoring | |
CN106304191A (en) | A kind of data receiver method based on cluster structured radio sensor network and device | |
CN107205154A (en) | A kind of radio multimedia sensor network compression of images acquisition method based on matrix completion | |
CN108924148A (en) | A kind of source signal collaborative compression perception data restoration methods | |
CN103781116A (en) | Data integration method for wireless sensor network based on distributed storage | |
CN102164395A (en) | Method for locally acquiring overall information of wireless sensor network based on compressed sensing | |
CN110062348B (en) | Wireless sensor network data collection method based on sparse random matrix | |
CN102355752B (en) | Wireless sensor network compressed sensing measurement matrix based on expander graph and reconfiguring method | |
CN105338602A (en) | Compressed data collection method based on virtual MIMO | |
CN114827931A (en) | WSN energy efficiency optimization routing method based on multi-agent reinforcement learning | |
CN104394565A (en) | Clustering multi-hop routing method based on maximum and minimum distance method | |
CN107743302B (en) | Rate allocation and routing combined optimization method based on compressed sensing | |
CN106211196B (en) | The harmonious property method of how unmanned sail body based on markov random file | |
CN108934029B (en) | Acceleration distributed optimization algorithm for perceptual big data reconstruction | |
AU2021101381A4 (en) | Energy-efficient prediction system for data aggregation in sensor network | |
CN110661566A (en) | Unmanned aerial vehicle cluster networking method and system adopting depth map embedding | |
CN104469797A (en) | Method for generating sequence prediction on basis of farmland wireless network intra-cluster data sparsity | |
CN104244358A (en) | Energy-saving routing strategy of wireless sensing network based on distributed compressed sensing (DCS) | |
Cui et al. | Information recovery via block compressed sensing in wireless sensor networks |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180504 |
|
RJ01 | Rejection of invention patent application after publication |